arXiv subjects
Publications and source records attributed to Tung Nguyen.
We show that if a graph $G$ admits a quasi-isometry $ϕ$ to a graph $H$ of bounded path-width, then we can assign a non-negative integer length to each edge of $H$, such that the same function $ϕ$ is a quasi-isometry to this weighted version of $H$, with error only an additive constant.
For finite graphs, path-width is an interesting and useful concept, but if we extend it to infinite graphs in the most obvious way (by making the indexing path infinite), it does not work nicely. The simplest extension that works nicely is to allow the indexing set to be any totally-ordered set, and then the corresponding decomposition is called a ``line-decomposition'', and the maximum bag size needed is called ``line-width''. In particular, the indexing set need not be a well-order; but the corresponding decomposition would be easier to use if it was. We show that if a graph has line-width at most $k$, it admits a well-ordered line-decomposition with width at most $2k$, and this is best possible.
Generative machine learning offers new opportunities to better understand complex Earth system dynamics. Recent diffusion-based methods address spectral biases and improve ensemble calibration in weather forecasting compared to deterministic methods, yet have so far proven difficult to scale stably at high resolutions. We introduce AERIS, a 1.3 to 80B parameter pixel-level Swin diffusion transformer to address this gap, and SWiPe, a generalizable technique that composes window parallelism with sequence and pipeline parallelism to shard window-based transformers without added communication cost or increased global batch size. On Aurora (10,080 nodes), AERIS sustains 10.21 ExaFLOPS (mixed precision) and a peak performance of 11.21 ExaFLOPS with $1 \times 1$ patch size on the 0.25° ERA5 dataset, achieving 95.5% weak scaling efficiency, and 81.6% strong scaling efficiency. AERIS outperforms the IFS ENS and remains stable on seasonal scales to 90 days, highlighting the potential of billion-parameter diffusion models for weather and climate prediction.
We confirm a conjecture of Fox, Pach, and Suk, that for every $d>0$, there exists $c>0$ such that every $n$-vertex graph of VC-dimension at most $d$ has a clique or stable set of size at least $n^c$. This implies that, in the language of model theory, every graph definable in NIP structures has a clique or anti-clique of polynomial size, settling a conjecture of Chernikov, Starchenko, and Thomas. Our result also implies that every two-colourable tournament satisfies the tournament version of the Erdős-Hajnal conjecture, which completes the verification of the conjecture for six-vertex tournaments. The result extends to uniform hypergraphs of bounded VC-dimension as well. The proof method uses the ultra-strong regularity lemma for graphs of bounded VC-dimension proved by Lovász and Szegedy and the method of iterative sparsification introduced by the authors in an earlier paper.
Menger's theorem says that, for $k\ge0$, if $S, T$ are sets of vertices in a graph $G$, then either there are $k + 1$ vertex-disjoint paths between $S$ and $T$, or there is a set X of at most $k$ vertices such that every $S$-$T$ path passes through $X$. The ``coarse Menger conjecture'' proposed a generalization of Menger's theorem for paths that are far apart: for all $k, c$ there exists $\ell$, such that for every graph $G$ and subsets $S, T \subset V (G)$, either there are $k + 1$ paths between $S$ and $T$, pairwise with distance more than $c$, or there is a set $X \subset V (G)$ of at most $k$ vertices such that every $S$-$T$ path has distance at most $\ell$ from $X$. This is known to be false, but may be true if $G$ is planar. Here we show that it is true if $G$ is planar and all vertices in $S \cup T$ are on the infinite region. In this case, we also obtain a linear-time algorithm to test for the existence of $k+ 1$ paths between $S$ and $T$, pairwise with distance more than $c$.
Regional weather forecasting is a critical problem for localized climate adaptation, disaster mitigation, and sustainable development. While machine learning has shown impressive progress in global weather forecasting, regional forecasting remains comparatively underexplored. Existing efforts often use different datasets and experimental setups, limiting fair comparison and reproducibility. We introduce IndiaWeatherBench, a comprehensive benchmark for data-driven regional weather forecasting focused on the Indian subcontinent. IndiaWeatherBench provides a curated dataset built from high-resolution regional reanalysis products, along with a suite of deterministic and probabilistic metrics to facilitate consistent training and evaluation. To establish strong baselines, we implement and evaluate a range of models across diverse architectures, including UNets, Transformers, and Graph-based networks, as well as different boundary conditioning strategies and training objectives. While focused on India, IndiaWeatherBench is easily extensible to other geographic regions. We open-source all raw and preprocessed datasets, model implementations, and evaluation pipelines to promote accessibility and future development. We hope IndiaWeatherBench will serve as a foundation for advancing regional weather forecasting research. Code is available at https://github.com/tung-nd/IndiaWeatherBench.
Coarse graph theory concerns finding 'coarse' analogues of graph theory theorems, replacing disjointness with being far apart. One of the most interesting open questions is to find a coarse analogue of Menger's theorem, which characterizes when there are $k$ vertex-disjoint paths between two given sets $S,T$ of vertices of a graph. We showed in an earlier paper that the most natural such analogue is false, but a weaker statement remained as a popular open question. Here we show that the weaker statement is also false. More exactly, suppose that $S,T$ are sets of vertices of a graph $G$, and there do not exist $k$ paths between $S,T$, pairwise at distance at least $c$. To make an analogue of Menger's theorem, one would like to prove that there must be a small set $X\subseteq V(G)$ such that every $S-T$ path of $G$ passes close to a member of $X$: but how small and how close? In view of Menger's theorem, one would hope for $|X|<k$ and 'close' some function of $k,c$ (and indeed, this was conjectured by Georgakopoulos and Papasoglu, and independently, by Albrechtsen, Huynh, Jacobs, Knappe and Wollan); but we showed that this is false, even if $c=3$ and $k=3$. Here we upgrade the counterexample: we show that, even if $c=k=3$, no pair of constants (for 'small' and 'close') work. For all $\ell, m$, there is a graph $G$ and $S,T\subseteq V(G)$, such that there do not exist three $S-T$ paths pairwise with distance at least three, and yet there is no $X$ with $|X|\le m$ such that every $S-T$ path passes within distance at most $\ell$ of $X$.
Deep learning-based prediction models for High-Level Synthesis (HLS) of hardware designs often struggle to generalize. In this paper, we study how to close the generalizability gap of these models through pretraining on synthetic data and introduce Iceberg, a synthetic data augmentation approach that expands both large language model (LLM)-generated programs and weak labels of unseen design configurations. Our weak label generation method is integrated with an in-context model architecture, enabling meta-learning from actual and proximate labels. Iceberg improves the geometric mean modeling accuracy by $86.4\%$ when adapt to six real-world applications with few-shot examples and achieves a $2.47\times$ and a $1.12\times$ better offline DSE performance when adapting to two different test datasets. Our open-sourced code is here: https://github.com/UCLA-VAST/iceberg
When $H$ is a forest, the Gyárfás-Sumner conjecture implies that every graph $G$ with no induced subgraph isomorphic to $H$ and with bounded clique number has a stable set of linear size. We cannot prove that, but we prove that every such graph $G$ has a stable set of size $|G|^{1-o(1)}$. If $H$ is not a forest, there need not be such a stable set. Second, we prove that when $H$ is a ``multibroom'', there {\em is} a stable set of linear size. As a consequence, we deduce that all multibrooms satisfy a ``fractional colouring'' version of the Gyárfás-Sumner conjecture. Finally, we discuss extensions of our results to the multicolour setting.
Foundation models have achieved remarkable success across video, image, and language domains. By scaling up the number of parameters and training datasets, these models acquire generalizable world knowledge and often surpass task-specific approaches. However, such progress has yet to extend to the domain of physics simulation. A primary bottleneck is data scarcity: while millions of images, videos, and textual resources are readily available on the internet, the largest physics simulation datasets contain only tens of thousands of samples. This data limitation hinders the use of large models, as overfitting becomes a major concern. As a result, physics applications typically rely on small models, which struggle with long-range prediction due to limited context understanding. Additionally, unlike images, videos, or text-which typically exhibit fixed granularity-physics datasets often vary drastically in scale, amplifying the challenges of scaling up multitask training. We introduce PhysiX, the first large-scale foundation model for physics simulation. PhysiX is a 4.5B parameter autoregressive generative model. It uses a discrete tokenizer to encode physical processes at different scales into a sequence of discrete tokens, and employs an autoregressive next-token prediction objective to model such processes in the token space. To mitigate the rounding error in the discretization process, PhysiX incorporates a specialized refinement module. Through extensive experiments, we show that PhysiX effectively addresses the data bottleneck, outperforming task-specific baselines under comparable settings as well as the previous absolute state-of-the-art approaches on The Well benchmark. Our results indicate that knowledge learned from natural videos can be successfully transferred to physics simulation, and that joint training across diverse simulation tasks enables synergistic learning.
Intracortical Brain-Computer Interfaces (iBCI) aim to decode behavior from neural population activity, enabling individuals with motor impairments to regain motor functions and communication abilities. A key challenge in long-term iBCI is the nonstationarity of neural recordings, where the composition and tuning profiles of the recorded populations are unstable across recording sessions. Existing methods attempt to address this issue by explicit alignment techniques; however, they rely on fixed neural identities and require test-time labels or parameter updates, limiting their generalization across sessions and imposing additional computational burden during deployment. In this work, we introduce SPINT - a Spatial Permutation-Invariant Neural Transformer framework for behavioral decoding that operates directly on unordered sets of neural units. Central to our approach is a novel context-dependent positional embedding scheme that dynamically infers unit-specific identities, enabling flexible generalization across recording sessions. SPINT supports inference on variable-size populations and allows few-shot, gradient-free adaptation using a small amount of unlabeled data from the test session. To further promote model robustness to population variability, we introduce dynamic channel dropout, a regularization method for iBCI that simulates shifts in population composition during training. We evaluate SPINT on three multi-session datasets from the FALCON Benchmark, covering continuous motor decoding tasks in human and non-human primates. SPINT demonstrates robust cross-session generalization, outperforming existing zero-shot and few-shot unsupervised baselines while eliminating the need for test-time alignment and fine-tuning. Our work contributes an initial step toward a robust and scalable neural decoding framework for long-term iBCI applications.
Let $\mathcal{F}$ be a set of subsets of a set $W$. When is there a tree $T$ with vertex set $W$ such that each member of $\mathcal{F}$ is the set of vertices of a subtree of $T$? It is necessary that $\mathcal{F}$ has the Helly property and the intersection graph of $\mathcal{F}$ is chordal. We will show that these two necessary conditions are together sufficient in the finite case, and more generally, they are sufficient if no element of $W$ belongs to infinitely many infinite sets in $\mathcal{F}$.
Recent advancements in mixed-modal generative have opened new avenues for developing unified biomedical assistants capable of analyzing biomedical images, answering complex questions about them, and generating multimodal patient reports. However, existing datasets face challenges such as small sizes, limited coverage of biomedical tasks and domains, and a reliance on narrow sources. To address these gaps, we present MedMax, a large-scale multimodal biomedical instruction-tuning dataset for mixed-modal foundation models. With 1.47 million instances, MedMax encompasses a diverse range of tasks, including interleaved image-text generation, biomedical image captioning and generation, visual chat, and report understanding. These tasks span knowledge across diverse biomedical domains, including radiology and histopathology, grounded in medical papers and YouTube videos. Subsequently, we fine-tune a mixed-modal foundation model on the MedMax dataset, achieving significant performance improvements: a 26% gain over the Chameleon model and an 18.3% improvement over GPT-4o across 12 downstream biomedical visual question-answering tasks. Finally, we introduce a unified evaluation suite for biomedical tasks to guide the development of mixed-modal biomedical AI assistants. The data, model, and code is available at https://mint-medmax.github.io/.
We prove that for every complete graph $K_t$, all graphs $G$ with no induced subgraph isomorphic to a subdivision of $K_t$ have a stable subset of size at least $|G|/{\rm polylog}|G|$. This is close to best possible, because for $t\ge 7$, not all such graphs $G$ have a stable set of linear size, even if $G$ is triangle-free.
How are Reddit communities responding to AI-generated content? We explored this question through a large-scale analysis of subreddit community rules and their change over time. We collected the metadata and community rules for over $300,000$ public subreddits and measured the prevalence of rules governing AI. We labeled subreddits and AI rules according to existing taxonomies from the HCI literature and a new taxonomy we developed specific to AI rules. While rules about AI are still relatively uncommon, the number of subreddits with these rules more than doubled over the course of a year. AI rules are more common in larger subreddits and communities focused on art or celebrity topics, and less common in those focused on social support. These rules often focus on AI images and evoke, as justification, concerns about quality and authenticity. Overall, our findings illustrate the emergence of varied concerns about AI, in different community contexts. Platform designers and HCI researchers should heed these concerns if they hope to encourage community self-determination in the age of generative AI. We make our datasets public to enable future large-scale studies of community self-governance.
Detecting and attributing temperature increases driven by climate change is crucial for understanding global warming and informing adaptation strategies. However, distinguishing human-induced climate signals from natural variability remains challenging for traditional detection and attribution (D&A) methods, which rely on identifying specific "fingerprints" -- spatial patterns expected to emerge from external forcings such as greenhouse gas emissions. Deep learning offers promise in discerning these complex patterns within expansive spatial datasets, yet the lack of standardized protocols has hindered consistent comparisons across studies. To address this gap, we introduce ClimDetect, a standardized dataset comprising 1.17M daily climate snapshots paired with target climate change indicator variables. The dataset is curated from both CMIP6 climate model simulations and real-world observation-assimilated reanalysis datasets (ERA5, JRA-3Q, and MERRA-2), and is designed to enhance model accuracy in detecting climate change signals. ClimDetect integrates various input and target variables used in previous research, ensuring comparability and consistency across studies. We also explore the application of vision transformers (ViT) to climate data -- a novel approach that, to our knowledge, has not been attempted before for climate change detection tasks. Our open-access data serve as a benchmark for advancing climate science by enabling end-to-end model development and evaluation. ClimDetect is publicly accessible via Hugging Face dataset repository at: https://huggingface.co/datasets/ClimDetect/ClimDetect.
Uncovering hidden topics from short texts is challenging for traditional and neural models due to data sparsity, which limits word co-occurrence patterns, and label sparsity, stemming from incomplete reconstruction targets. Although data aggregation offers a potential solution, existing neural topic models often overlook it due to time complexity, poor aggregation quality, and difficulty in inferring topic proportions for individual documents. In this paper, we propose a novel model, GloCOM (Global Clustering COntexts for Topic Models), which addresses these challenges by constructing aggregated global clustering contexts for short documents, leveraging text embeddings from pre-trained language models. GloCOM can infer both global topic distributions for clustering contexts and local distributions for individual short texts. Additionally, the model incorporates these global contexts to augment the reconstruction loss, effectively handling the label sparsity issue. Extensive experiments on short text datasets show that our approach outperforms other state-of-the-art models in both topic quality and document representations.
Menger's well-known theorem from 1927 characterizes when it is possible to find $k$ vertex-disjoint paths between two sets of vertices in a graph $G$. Recently, Georgakopoulos and Papasoglu and, independently, Albrechtsen, Huynh, Jacobs, Knappe and Wollan conjectured a coarse analogue of Menger's theorem, when the $k$ paths are required to be pairwise at some distance at least $d$. The result is known for $k\le 2$, but we will show that it is false for all $k\ge 3$, even if $G$ is constrained to have maximum degree at most three. We also give a simpler proof of the result when $k=2$.